pull down to refresh
27 sats \ 6 replies \ @Undisciplined 14 Feb \ on: Bitcoin Boosted by Pornhub (SSRN, Will Luther) econ
On the common trends point, it doesn't really matter if the basket containing bitcoin was trending the same way as the other basket (in the sense people would usually think of). It's easy enough to include a time trend in the regression to account for different trend lines.
The point is that the debanking date is only relevant to the bitcoin basket and was not relevant to anything related to the other basket. Now, that assumption is very problematic, since crypto prices are not likely to be independent from each other.
One robustness check to run is just doing the same regression with a bunch of other random dates to see if the result is still there.
Alrightsy, alrightsy. Above my metrics paygrade... It did have my alarm bells go off when he merged all those shitcoins w BTC to create one indicator. Not lovely
reply
I haven't looked into his methods, but that wouldn't be the thing that gave me pause.
reply
I think if he really wants to prove his point while also addressing @denlillaapan's concerns, he'd create a basket of everything Phub accepts, excluding bitcoin, and compare it to the basket of shitcoins that Phub didn't accept.
reply
then nobody would read the paper, haha.
Nobody cares about shitcoins; bitcoin makes it a juicy paper!
reply
He should've made 3 plots:
- Bitcoin
- shitcoins accepted by Phub, excluding bitcoin)
- shitcoins not accepted by Phub
reply
I agree, since the headline takeaway is about the effect on Bitcoin, not a bundle of random crap that includes bitcoin.
reply